Hi all,

My s90 (101c) intermittently fails to boot properly; it will just power off again early in the boot process. I'm not sure if this means chdk has crashed. I spent some time bisecting the svn revision that introduced this behaviour, and it turns out to be r972:

------------------------------------------------------------------------
r972 | reyalp | 2010-11-15 09:02:15 +0800 (Mon, 15 Nov 2010) | 1 line

remove spurious "conf_change_script_file" from text reader conf code. thx emlyn and ultimA
------------------------------------------------------------------------

Which looks trivial:

Index: core/conf.c
===================================================================
--- core/conf.c	(revision 971)
+++ core/conf.c	(revision 972)
@@ -181,7 +181,7 @@
     CONF_INFO( 35, conf.raw_in_dir,             CONF_DEF_VALUE, i:0, NULL),
     CONF_INFO( 36, conf.raw_prefix,             CONF_DEF_VALUE, i:RAW_PREFIX_CRW, NULL),
     CONF_INFO( 37, conf.raw_ext,                CONF_DEF_VALUE, i😁EFAULT_RAW_EXT, NULL),
-    CONF_INFO( 38, conf.reader_file,            CONF_DEF_PTR,   ptr:"A/README.TXT", conf_change_script_file),
+    CONF_INFO( 38, conf.reader_file,            CONF_DEF_PTR,   ptr:"A/README.TXT", NULL),
     CONF_INFO( 39, conf.reader_pos,             CONF_DEF_VALUE, i:0, NULL),
     CONF_INFO( 40, conf.sokoban_level,          CONF_DEF_VALUE, i:0, NULL),
     CONF_INFO( 41, conf.show_clock,             CONF_DEF_VALUE, i:2, NULL),

Reverting this commit makes the camera work as per normal. Since nobody else has a problem with this, could it be a problem with my flashcard files? It may be that I don't have a readme.txt in the proper location.

The use of conf_change_script_file seems incorrect (and the commit correct), but could it be that thing don't work well if script_load (called by conf_change_script_file) is called much later in the conf handling process (entry 38 vs 194)?

Cheers,

Rob